Multi-Agent Debate Strategies: Survey, Taxonomy, and Challenges
Source
arxiv.org
Author
Quim Motger, Marc Oriol, Jordi Marco, Xavier Franch
Date
Why it matters
multi-agentUsing several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.Full definition → debate setups have quietly standardized on static fully connected topologies, verbatim exchange, short-term memory and voting — by habit, not evidence. The design space is about a dozen interacting choices most papers leave unstated, which makes cross-study comparison unreliable.
Terms in this piece · Glossary
multi-agent — Using several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.
articleBeyond Symmetric Agents: Cognitive Diversity and Multi-Agent Debate in Small Language ModelsLeonardo Ferreira, Gardenia Liu, Kaden Zheng
articleEven More Deception: Objective Misalignment in Mixed-Motive LLM Multi-Agent SystemsMarylou Fauchard, Florian Carichon, Margarida Carvalho, Golnoosh Farnadi
postSheaf-ADMM: agents with partial views negotiate to a correct global answerSakana AI